Merge pull request #147 from sjackman/quast

Add quast

Author: ODiogoSilva

flowcraft/generator/components/assembly_processing.py

@@ -229,3 +229,38 @@ def __init__(self, **kwargs):
                 "version": "1.22.0-1"
             }
         }
+
+class Quast(Process):
+    """Assess assembly quality using QUAST
+
+    This process is set with:
+
+        - ``input_type``: assembly
+        - ``output_type``: tsv
+        - ``ptype``: post_assembly
+
+    """
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        self.input_type = "fasta"
+        self.output_type = "tsv"
+
+        self.params = {
+            "reference": {
+                "default": "null",
+                "description": "Compare the assembly to this reference genome"
+            },
+            "genomeSizeBp": {
+                "default": "null",
+                "description": "Expected genome size (bp)"
+            },
+        }
+
+        self.directives = {
+            "quast": {
+                "container": "quay.io/biocontainers/quast",
+                "version": "5.0.0--py27pl526ha92aebf_1"
+            }
+        }

flowcraft/generator/engine.py

@@ -91,6 +91,7 @@
         "process_spades": ap.ProcessSpades,
         "progressive_mauve":alignment.ProgressiveMauve,
         #"prokka": annotation.Prokka,
+        "quast": ap.Quast,
         "raxml": phylogeny.Raxml,
         "reads_download": downloads.DownloadReads,
         "remove_host": meta.RemoveHost,

flowcraft/generator/templates/quast.nf

@@ -0,0 +1,48 @@
+if (params.reference{{param_id}} == null && params.genomeSizeBp{{param_id}} == null)
+    exit 1, "Specify at least one of reference or genomeSizeBp"
+if (params.reference{{param_id}} != null && params.genomeSizeBp{{param_id}} != null)
+    exit 1, "Specify only one of reference or genomeSizeBp"
+
+if (params.reference{{param_id}} != null) {
+    process quast_{{pid}} {
+        {% include "post.txt" ignore missing %}
+
+        tag { sample_id }
+        publishDir "results/assembly/quast_{{pid}}/$sample_id", pattern: "*.tsv"
+        publishDir "reports/assembly/quast_{{pid}}/$sample_id"
+
+        input:
+        set sample_id, file(assembly) from {{input_channel}}
+        file reference from Channel.fromPath(params.reference{{param_id}})
+
+        output:
+        file "*"
+        {% with task_name="quast" %}
+        {%- include "compiler_channels.txt" ignore missing -%}
+        {% endwith %}
+
+        script:
+        "/usr/bin/time -v quast -o . -r $reference -s $assembly -l $sample_id -t $task.cpus >> .command.log 2>&1"
+    }
+} else if (params.genomeSizeBp{{param_id}} != null) {
+    process quast_{{pid}} {
+        {% include "post.txt" ignore missing %}
+
+        tag { sample_id }
+        publishDir "results/assembly/quast_{{pid}}/$sample_id", pattern: "*.tsv"
+        publishDir "reports/assembly/quast_{{pid}}/$sample_id"
+
+        input:
+        set sample_id, file(assembly) from {{input_channel}}
+        val genomeSizeBp from Channel.value(params.genomeSizeBp{{param_id}})
+
+        output:
+        file "*"
+        {% with task_name="quast" %}
+        {%- include "compiler_channels.txt" ignore missing -%}
+        {% endwith %}
+
+        script:
+        "/usr/bin/time -v quast -o . --est-ref-size=$genomeSizeBp -s $assembly -l $sample_id -t $task.cpus >> .command.log 2>&1"
+    }
+}